Home

R-integration package - Agilent Technologies

Contents

1. sess 23 3 4 5 Running the External Program eeeeeeeeeeeeeeetstee entente tne ete oannes tte deae ee 26 3 4 6 Using the Graphical User Interface for R sees 27 4 The GeneSprng R Package ee ette SER ent IR Ree NE npe ees 28 4 Installing the GeneSpring package 2 2 28 4 2 The GeneS pny TUNCUONS RR 33 5 Appendix A Complete 34 5 1 BOX Ob EI 34 5 2 34 5 3 SAP MULCH ess 35 54 1 36 5 5 101 37 5 6 Power OF ANOVA 38 5 7 Q valtic With Plots 39 5 8 Z score normalizatiofi 41 2 Introduction and summary 2 1 Introduction Gene Expression analysis has become a very mature field and there are many different analyses that can be performed using gene expression data Many scientists develop their own algorithms that will help them understand their data and allow them to answer their scientific questions Many of the custom analyses are developed in specialized packages like R SAS Matlab JMP etc and many users would like to be able to integrate their specialized algorith
2. 2 eo E z Expression Time hours 124 48 96 144 From the External Program Inspector window select the Edit Program Details button to edit the external program definition GeneSpring GX integration package 17 18 External Program Inspector Program Name Author s your name Research Group your group Anthonius de Boer kfith 5342 Sat Aug 30 0 02 56 PDT 2003 Full Development 6 1 Directory Location C Program Files SiliconGenetics GeneSpring data Programs Filter CV programdef I Program Details Type of Executable Executable Program Command GS exec R bat cv filter R Inputto External Program Output from External Program Experiment Condition Statistics Gene List With Numbers Change the Command Line section By default the path is opt SiliconGenetics GeneSpring data GS exec H sh Edit External Program Details Program Executable E External Program Java Class C HTTP Command Line ppt SiiconGenetics GeneSpringrdata S exec R bat cv filter Browse Inputs outputs Delimiters Arguments Input from GeneSpring to External Program Add Input Experiment Condition Statistics Debug Input If GeneSpring GX is not located in the opt directory change the path accordingly For example if your login name for UNIX is peter and your home directory is in home peter and you installed GeneSpring
3. EEEE EEEE EEE H AE EEEE EEEE EEEE EEEE EEE EEEE EEEE EEE EEEE EEEE EEEE EEEE EEE EEEE EEEE EE E library graphics library stats library utils Hn Original code by Storey starts here Version 1 1 Updated June 2003 These functions estimate the q values for a given set of p values The statistical methodology mainly comes from Storey JD 2002 A direct approach to false discovery rates Journal of the Royal Statistical Society Series B 64 479 498 5 http faculty washington edu jstorey qvalue for more info 11 functions were written by John D Storey Copyright 2002 2003 by John D Storey 11 rights are reserved and no responsibility is assumed for mistakes in or caused by the program qvalue lt function p lambda seq 0 0 95 0 05 piO0 meth smoother fdr level NULL robust F fInput a vector of p values only necessary input fdr level a level at which to control the FDR optional lambda the value of the tuning parameter to estimate 10 optional pi0 method either smoother or bootstrap the method for automatically choosing tuning parameter the estimation of pi0 the proportion of true null hypotheses robust an indicator of whether it is desired to make the estimate more robust for small p values a direct finite sample estimate of pFDR optional Output call gives the function call 0 an estimate of the proporti
4. lt sour If it is not convenient to read the STDIN and write to STDOUT you can also redirect the input output to a file and read that file in stead The program can either be a command line program a JAVA program or a HTTP POST request but the principle is the same GeneSpring GX will send a stream of data to the program which will perform the analysis and send back a stream of information to GeneSpring GX 3 2 Setup Agilent has provided an R integration package for GeneSpring GX that can install all the necessary components on your computer The installation package can be simply dropped onto GeneSpring GX version 6 1 which will install the R package and a number of examples which will help you create your own library of R scripts that will expand the capability of GeneSpring GX For those users that already have a copy of R we have also provided a version of the installer that only contains the GeneSpring GX specific part of the R integration package This version may require the user to edit two files to indicate the path to the R program To install the R integration package perform the following steps 1 If you do not already have installed and will not install the complete Windows version gs r winall zip download and install R from http y or n us r project org 2 If you will not install the complete Windows version gs r winall zip which contains the complete R package you need to install the R GeneS
5. http www bioconductor org 3 2 2 3 Installing GeneSpring GX R library The GeneSpring GX R integration package uses a special library called GeneSpring This library performs most of the actual conversion functions and is required for the use of the GeneSpring GX R integration package This library is now part of the BioConductor release 1 6 and later For information about installation of the GeneSpring R library see Chapter 4 The R library is called GeneSpring GX allowing the library to be loaded in an session by the command GeneSpring GX R integration package library GeneSpring Note the capitalization See Section 4 1 for more information about the functions in the library GeneSpring GX or check the vignette in R 3 2 2 4 Installing GeneSpring GX R Integration package into GeneSpring GX Starting with version 6 1 GeneSpring GX can install any GeneSpring GX related item with a special Drag and Drop interface that will make it easy to install new functionality and data into GeneSpring GX With the Drag and Drop ZIP technology you will be able to install new pathways new genomes new physical maps new scripts and external programs and Agilent will use the new Drag and Drop interface to continue to enhance the functionality of GeneSpring GX via this interface The GeneSpring GX R integration package is the one of the first examples of this technology If the ZIP file is not already present o
6. lt the script to be executed is copied into a special file called Rprofile This file is automatically loaded and executed upon startup of Rgui and this will ensure that our script is executed NOTE The R script is copied into the file Rprofile in the GeneSpring GX data directory and any Rprofile file that was already there will be overwritten If your implementation requires your own Rprofile file make appropriate changes to the script or moving your Rprofile changes to a site profile UNIX Mac OS X GS exec HR sh The GS exec H sh script that is provided on the UNIX systems R integration packages performs basically the same functionality as the GS exec R bat script for Windows described above The script is given below bin bash Get the R script file name Rscript 1 shift Set the script parameters as environment variables while 4 gt 0 do export 1 shift done Take the output of GeneSpring and put it in a file for easier access cat gt GS_R_in txt Run R Let s hope it is in the path If not edit this R CMD BATCH vanilla slave Rscript Put the results back into stdout cat GS R out txt Clean up after ourselves rm GS out txt GS R in txt The script takes the R script as the first command line argument and saves this into a variable after which it sets the External Program s parameters as environment variables The script can take an unlimited number of parameters The data from
7. IF 1 GOTO Continue set 1 2 shift shift GOTO Loop Continue cat exe GS R in txt rw2010 bin Rterm exe no restore slave lt R_SCRIPT gt Rterm output txt cat exe GS R out txt del GS R in txt GS R out txt Rterm output If you would wanted to use the bat file on the Windows command line you would type something like GS exec R bat cv filter R CV cutoff 100 Matching conditions 1 INPUT txt OUTPUT txt This would execute the R script cv filter R located in the GeneSpring GX data directory with the R program using the parameter CV cutoff set to 100 and the parameter Matching conditions set to 1 3 4 1 1 GS exec R bat file explained A small explanation of each of the lines in the batch file is given below echo off This line instructs the script to not type the commands on the Standard Output since this would interfere with the data parsing within GeneSpring GX later set R SCRIPT 1 This line saves the filename of the R script to be executed in the environment variable R SCRIPT shift Loop IF 1 GOTO Continue set 1 2 shift shift GOTO Loop Continue This set of lines assign the parameters that need to be passed to the R script GeneSpring GX will provide the script arguments and passes them on to the script GS exec R bat as command line parameters as key value pairs In this section the script loops through each set of key value pairs that are provided on the command l
8. Packages Load package R Console Set CRAN mirror Copyright Install package s Computing Version 2 1 0 Update packages R is free soft Install package s from local zip files You are welcom ditions Type license or licence for distribution details Natural language support but running in an English locale R is a collaborative project with many contributors Type contributors for more information and citation on how to cite or R packages in publications demo for some demos help for on line help or help start for HTML browser interface to help Type 9 to quit R gt utils setRepositories gt utils menuInstallPkgs Error in install packages NULL libPaths 1 dependencies TRUE type typ I no packages were specified gt R Language and Environment A list of available repositories will be shown Select the Bioconductor option and press OK GeneSpring GX R integration package 29 Repositories Biocondu Or Load package Set CRAN mirror Select repositories Install package R Console R Copyright Version 2 1 0 Update packages Computing R is free soft Install package s from local zip files You are welco Type license or licence for distribution details Natural language support but running in an English locale R is a collaborative project with many contribut
9. pid rep 0 length lambda for i in 1 length lambda 10 1 lt mean p2 gt lambda i 1 lambda i spi0 lt smooth spline lambda pi0 df 3 GeneSpring GX R integration package 100 lt round qobj pi0 3 plot lambda pi0 xlab expression lambda ylab expression hat pi 0 lambda pch mtext substitute hat pi 0 that list that pi00 lines spi0 plot p2 q2 rng q2 q2 lt rng type 1 xlab p value ylab q value plot q2 g2 2rng 1 sum q2 rng type l xlab q value cut off ylab significant tests plot 1 sum q2 lt rng q2 q2 lt rng 1 sum q2 2rng type l xlab significant tests ylab expected false positives par mfrow c 1 1 qwrite function qobj filename my qvalue results txt fInput qobj a q value object returned by the qvalue function filename the name of the file where the results are written Output file sent to filename with the following First row the function call used to produce the estimates Second row the estimate of the proportion of false positives 10 Third row and below the p values 1st column and the estimated q values 2nd column cat c Function call deparse qobj call file filename append F cat c Estimate of the overall proportion of false positives pi0 gobj pi0 n file filename append T for i in l length qobj qval cat c qobj pval i qobj
10. GeneSpring GX integration package 15 3 3 Post installation modifications There are a few modifications that may need to be performed in order to ensure that the R scripts can be executed successfully For each of the supported platforms there are a few possible changes and they are provided in the next sections 3 3 1 WINDOWS 3 3 1 1 Change the location of the program OPTIONAL The script GS exec R bat and GS exec R GUl bat assume that the R program is installed in the GeneSpring GX data directory This is where the full GeneSpring GX R integration package will install version 2 1 0 of the R program If R was not installed in this way and the user would like to use stored in a different location edit the GS exec H bat and GS exec R GUl bat scripts These scripts are located in the data directory of the main GeneSpring GX folder When GeneSpring GX was installed in the default location the path to the GS exec R bat script would be C Program Files SiliconGenetics GeneSpring data GS_exec_R bat Change the following line to reflect the location of the R program rw2010 bin Rterm exe no restore slave R_SCRIPT gt Rterm output txt To something like YOUR PATH HERE Rterm exe no restore slave lt R SCRIPTS gt Rterm output txt The bat file uses the program called Rterm exe that is located in the bin directory of the main R installation folder NOTE When your R program is located in a directory that h
11. cran r project org manuals html The R scripts should be stored in the GeneSpring GX data directory so they can be easily referenced in the External Program definition 3 4 4 Create the External Program definition in GeneSpring GX Once all the files are created and stored in their correct place GeneSpring GX needs to be instructed to be able to use the scripts and we will define them as external programs To configure GeneSpring GX to use the external programs open GeneSpring GX and go to the menu File gt New External Program This will open up a dialog box as displayed below amp New External Program H External Programs Icon Browse O winsight Program Executable External Program Java Class C HTTP Command Line Browse Inputs outputs Delimiters Arguments Input from GeneSpring to External Program Add Input No Data Debug Input Help Name Start by giving the new External Program definition a name like Filter on CV since we will be using the example from Appendix B Folder The external program can be placed in its own folder and you could either type the name of a new or existing folder or select the folder from the navigator on the left Icon If you want you can also assign an icon to the External Program but in most cases leaving this blank is fine With the GeneSpring GX R Integration package we have provided a special icon that could b
12. lt min pi0 1 1 10 lt 0 print ERROR The estimated 10 lt 0 Check that you have valid p values or use another lambda meth return 0 estimated q values calculated here u lt order v lt rank qvalue lt pid m p v f robust qvalue lt pi0 m p v 1 1 p m qvalue u m lt min qvalue u m 1 for i in m 1 1 qvalue u i lt min qvalue u i qvalue u it 1 1 The results are returned if lis null fdr level retval lt list callematch call 10 10 qvalues qvalue pvalues p significant qvalue lt fdr level lambda lambda else retval lt list call match call 10 10 qvalues qvalue pvalues p lambda lambda class retval qvalue return retval qplot function qobj rng 0 1 Input qobj a q value object returned by the qvalue function rng the range of q values to be plotted optional Output Four plots Upper left pi0 hat lambda versus lambda with a smoother Upper right q values versus p values Lower left number of significant tests per each q value cut off Lower right number of expected false positives versus number of significant tests q2 lt 1 order qobj pval if min q2 gt rng rng quantile q2 0 1 p2 lt gobj pval order qobj pval par mfrow c 2 2 lambda lt gob j lambda if length lambda 1 lambda lt seq 0 max 0 95 lambda 0 05
13. my fdr adjust function lpe result adjp BH if adjp BH adjp BY x location lt grep x names lpe result y location lt grep y names lpe result x lt lpe result x location y lt lpe result y location pnorm diff lt pnorm lpe result median diff mean 0 sd lpe result pooled std dev p out lt 2 apply cbind pnorm diff 1 pnorm diff 1 min p adj mt rawp2adjp p out proc adjp data out lt data frame x x median 1 lpe result median 1 std dev 1 lpe result std dev l y y median 2 lpe result median 2 std dev 2 lpe result std dev 2 median diff lpe result median diff pooled std dev lpe result pooled std dev abs z stats abs lpe result z stats p adj p adj return data out Get environment variables Param Sys getenv c ExperimentParameter Conditionl Condition2 PvalueCutoff MulTestingCorrection ExperimentParameter as character Param 1 Conditionl as character Param 2 Condition2 as character Param 3 PvalueCutoff if as numeric Param 4 0 0 else if as numeric Param 4 gt 1 1 else as numeric Param 4 MulTestingCorrection ifelse is na grep as character Param 5 c BH BY resamp none no 1 BH as character Param 5 Read the Experiment Interpretation The experiment gs int GSload exp GeneSpring GX R integration package Get rid of the missing v
14. Depending on the version of the zip file you downloaded the text may be slightly different from the one displayed above The installation will take a few minutes and after the installation is complete another dialog box will be shown informing you that the installation was successful and GeneSpring GX is ready to be shut down Upgrade Complete Upgrade complete Click Ok to quit GeneSpring Occasionally it is possible that GeneSpring GX will first display a message that certain files will be overwritten and you can still abort the installation procedure at this point by clicking For other Drag and Drop files in the future like Pathways etc it may make a difference which Genome you drop the ZIP file on GeneSpring GX integration package 7 the No button if you feel that you would like to double check that the files you are about the overwrite can indeed be updated Caution This upgrade will overwrite some existing files which were most recently installed on Nov 9 2003 11 21 17 Do you wish to proceed The first time you install the GeneSpring GX R Integration package you should NOT see this message since all files are supposed to be new 3 2 2 5 New External Programs Once you have installed the GeneSpring GX R Integration package the next time you start up GeneSpring GX you should see a number of new items in the External Programs and the Scripts section The External programs se
15. Experiment interpretation and calculates the power anova for each gene Author Thon de Boer Agilent Technologies Redwood city CA Last Update 23 Mar 2005 L Version 2 0 1 Environment variable SignificanceRequired The desired significance for the power calculation ERE ERE ERE HERE RRR RE HEE RR ERE PEE RRR HEH EEEE EEEE EE EEE EEEE EEEE EEEE EE EEEE EEEE EH AE HEE HSE EEEE EEE EEEE EE Ed Prepare GeneSpring environment library graphics library stats library utils library GeneSpring Read the Experiment Interpretation gs int GSload int Get the Parameters and do some sanity checking Param Sys getenv c SignificanceRequired SignificanceRequired if Param 1 lt 0 0 else if Param 1 gt 1 1 else as numeric Param 1 Create list of Variances BETWEEN the groups the variances of all the means of the groups Column 1 of exp VarBetween apply gs int a nor 1 var GeneSpring GX R integration package Create list of Variances WITHIN the groups the median of all the SD data columns and convert it to variances Column 2 of exp VarWithin apply gs int sd nor 1 median 2 Create a list of the number of replicates for each gene the median of all the numbers of replicates Column 3 of exp numReps lt apply gs intG8n nor 1 median Calculate the power for each gene powerValues power anova test groups
16. FE AE EH HE HEE ERE HERE HEHEHE HAE EEE PEER dE dE dE db de db d db GeneSpring environment library graphics library stats library utils library GeneSpring Read the Experiment gs int GSload exp 2 score function definition zscore lt function 1 This function takes a list and returns the Z score for each of the members of the list The mean and SDev are calculated over the members of the list Input list of numbers Outout list of dim list with the Z score for each of the members GeneSpring GX integration package 41 42 s sd 1 na rm TRUE m mean 1 na rm TRUE return 1 m s End of Z score function definition Calculate the Z scores a nor new lt t apply gs int a nor 1 zscore a new GeneSpring experiment with just the Z score data We could not try to convert the COntrol values to retain the original RAW values since GeneSpring cannot deal with negative Control values The experiment name gets suffixed with Z score and the parameters are copied newGSint lt GSint a nor a nor new expName paste gs int expName Z score expparam gs int expparam genenames gs int genenames Save the experiment GSsave exp newGSint GeneSpring GX R integration package
17. GX with the default settings you should change the path to home peter SiliconGenetics GeneSpring data GS_exec_R sh Make the changes in all 7 external programs that are provided as examples GeneSpring GX integration package 3 3 2 2 Change the permissions of the GS exec R sh script REQUIRED Since the GS exec R sh needs to be executed the execute permissions need to be set for the program using the UNIX chmod command The GS exec R sh script is located in the data directory of the main GeneSpring GX installation chmod a x GS exec R sh This will ensure that the script can be executed by GeneSpring GX 3 3 2 3 Change the location of the program OPTIONAL The script GS exec R sh assumes that the R program is installed in a directory that is in the user s path If the R program is not in the path edit the path for the user or edit the GS exec H sh script Change the following line to reflect the location of the R program CMD BATCH vanilla slave lt S Rscript To something like YOUR PATH HERE R CMD BATCH vanilla slave lt Rscript 3 3 2 4 Non default data directory location OPTIONAL If the preference setting for the location of the GeneSpring GX data directory has been changed in the past the scripts will not work You will need to move the GS exec F sh file and the R scripts to the original data directory in order for the scripts to run successfully If you have changed your data directory set
18. GeneSpring GX is saved into a file called GS since it is a little easier to deal with the data in files that it is with data in the stdin stream NOTE The R program is assumed to be in the command path of the user running the script If this is not guaranteed edit the line to indicate the complete path to the R program as described in section 3 3 2 3 After running the actual script the output is copied from a file onto the stdout stream and the temporary files are cleaned up Create the R script In this step you will write or install the R script that performs the actual function you would like to integrate into GeneSpring GX The R script should typically consist of three parts 1 Readin the data and the parameters 2 Perform the analysis 3 Save the results in a file or the Standard Output stream Most of the time you would only need to have to worry about the second point since we have provided a suite of standard R functions that will perform some of the tedious but important tasks of loading the data and parsing it out in data objects that can be manipulated in R We have provided an package called GeneSpring that is part of BioConductor version 1 4 0 and later and can be used to load and parse the data See Appendix A for a description of the GeneSpring R package GeneSpring integration package For more information about writing R scripts refer to the User Manual for R at http
19. any character to be used as a systematic name NOTE R scripts written to use GeneSpring GX library 1 0 x ARE NOT COMPATIBLE with this new version of the GeneSpring GxX library 5 All of the scripts were updated to work with R version 2 1 The contents of certain libraries have been changed in R version 2 1 0 and the example R scripts have all been updated to work with R version 2 1 Summary This document contains a description of how be able to run R scripts from within the GeneSpring GX Interface The document describes the installation of the R integration package provided by Agilent and contains numerous examples of functional R scripts that perform useful analysis GeneSpring integration package This document is also available from our website at http www chem agilent com cag bsp SiG Downloads pdf gs r doc 3 The GeneSpring GX External Program Interface 3 1 How the EPI works The GeneSpring GX External Program Interface EPI will allow GeneSpring GX to start an external program and provide this program with the data to be analyzed When the program is finished with its analysis the results will be passed back to GeneSpring GX which will present the user with the results in the familiar GeneSpring GX Interface The only thing the external program needs to do is to read the standard input stream STDIN and write the results to the standard output stream STDOUT and GeneSpring GX will do the rest
20. c gs int8numConditions n numReps numReps gt 0 between var VarBetween numReps gt 0 within var VarWithin numReps gt 0 sig level SignificanceRequired power NULL Create the two genelists genelist lt data frame genes gs int8genenames numReps gt 0 power as numeric powerValues power Write the resulting Gene List GSsave genelist genelist 5 7 Q value with plots This is a complete example of an R script This program will take an as input a list of P values and calculate a list of Q values and filter the Q values based on the provided Qvalue cutoff The program will also produce four plots allowing the user to determine appropriate Q value cutoff values EEEE EE EEEE EEE EEEE EEEE EEE EEE EEE EE EEEE qvalue R Program that reads GeneSpring gene list with P values as associated values and returns a gene list with values as associated values that pass the cutoff Version 1 1 1 Author Thon de Boer Agilent Technologies SiliconGenetics Redwood city CA Last Update 24 Jun 2004 4 Environment variable Q cutoff Value for the cutoff of the Q value Environment variable bootsrap Indicates if the bootstrap method should be used Possible values bootstrap smoother Environment variable robust Indicates whether it is desirable to make the estimate more robust Possible values yes no Default
21. int Get the Parameters and do some sanity checking Param Sys getenv c CV cutoff Matching conditions Data to use Turn parameter into variables CVcutoff if Param 1 0 0 else as numeric Param 1 Nconditions if Param 2 lt 0 1 else as numeric Param 2 firstLetter lt substr toString Param 3 1 2 DataToUse lt ifelse firstLetter r firstLetter R ifelse firstLetter n firstLetter N R N R Create table of CV values using data described by DataToUse If the data is not there we ll get into trouble but we ll just return an empty genelist CVvalues lt if DataToUse N 100 gs int sd nor gs int a nor else 100 gs int sd raw gs int a raw a table of all those values that pass the cutoff and check for those entries that don t have a SD value and have a NaN value CVpass ifelse is na CVvalues lt CVcutoff FALSE CVvalues lt CVcutoff a list of those genes that pass the Nconditions setting CVnumber lt apply CVpass 1 sum CVok lt if Nconditions gt 1 CVnumber gt Nconditions else CVnumber gs int numConditions gt Nconditions genelist lt data frame genes gs int genenames as logical CVok pass as numeric CVnumber as logical CVok Write the resulting Gene List GSsave genelist genelist 5 3 gapFilter This script implements the BioConductor gapFilter genefilter
22. 004 atBurkitt lymphoma receptor 1 GTP binding protein 1005 at dual specificity phosphatase 1 1006 atmatrix metalloproteinase 10 stromelysin 2 1007 s atHuman receptor tyrosine kinase DDR gene complete cds 1008 f at PKR P1 kinase p68 kinase Human interferon inducible Rt 1009_at histidine triad nucleotide binding protein 1 101_at dual specificity tyrosine Y phosphorylation regulated kinase 4 _ Similar lists Show as List c Show as Navigator P value List Name 0 0 Subgroup nm Subgroup pm axis 0 0 Subgroup nm Subgroup pm M y axis 0 0 CV lt 100 atleast 5 conditions on GL all 0 0 Qvalue 0 05 on all genes 0 0 CV 50 in all conditions on GL all genes 0 0 Q value lt 0 05 on CV lt 50 in all conditions on t 3 235029E 5 like 34344 at IKBKAP 0 95 Save Cancel From this point the R script is successfully integrated into GeneSpring GX 3 4 6 Using the Graphical User Interface for R In the previous example we used the Rterm exe program from the R package which will execute the R script without providing any Graphical Interaction or feedback It is possible to use the Graphical User Interface for the R package which will allow us to display any plots or other graphical output that does not contain any data that can be stored in GeneSpring GX In order to be able to use the GUI interface to R the Rgui exe program should be used in stead of the Rterm exe program to execute the R script N
23. 3 2 2 6 7 Q value from 1 way ANOVA results Script calculates a 1 way ANOVA on the selected Gene List and Experimental Parameter It will calculate the Q value and will return a Gene List of those genes that have a Q value below the cutoff value Qcutoff The Q value is a P value that is corrected for the multiple tests that are performed in the ANOVA It computes the multiple testing correction factor to limit the false discovery rate For a given gene i with P value pi new P value Q value becomes pie The statistical methodology mainly comes from Storey JD 2002 A direct approach to false discovery rates Journal of the Royal Statistical Society Series B 64 479 498 See http faculty washington edu jstorey qvalue for more info Input 1 Starting Gene List 2 Experiment condition The Experiment condition should have at least 2 conditions Knobs 1 Groups Specification The name of the Experimental parameter to be used in the 1 way ANOVA The parameter should be entered using the GeneSpring GX specific format Experimental parameter 2 cutoff The cutoff value for the Q value Only genes with value at or below this value will be returned 3 bootstrap Parameter will determine the method for automatically tuning the parameter in the estimation of pio the proportion of true null hypotheses Set to yes or y for choosing the bootstrap method Use no or n for using the smoother metho
24. Conductor Only when you did not install the complete R integration package would you require installing the GeneSpring R library and a short description about the installation of the GeneSpring package is given below Installing the GeneSpring package To install the GeneSpring R library open the Graphical User interface for the program GeneSpring GX R integration package Ele Edit Misc Packages Windows el R Console Copyright 2005 The Foundation for Statistical Computing Version 2 1 0 2005 04 18 ISBN 3 900051 07 0 R is free software and comes with ABSOLUTELY NO WARRANTY You are welcome to redistribute it under certain conditions Type license or licence for distribution details Natural language support but running in an English locale R is a collaborative project with many contributors Type contributors for more information and citation on how to cite R or R packages in publications Type demo for some demos help for on line help or help start for HTML browser interface to help Type q to quit R gt utils setRepositories gt utils menuInstallPkgs Error in install packages NULL libPaths 1 dependencies TRUE type typ no packages were specified i Language and Environment From the menu Packages ensure that the BioConductor repository is activated by choosing Select repositories from the Packages menu
25. G 2001 Significance analysis of microarrays applied to the ionizing radiation response _PNAS_ 98 5116 5121 Version 2 0 0 Author Thon de Boer Agilent Technologies Redwood city CA Last Update 18 Feb 2005 Environment variable ExperimentParameter Name of the experimental parameter that define the groups Environment variable Conditionl Value of the eperimental parameter Condition defining group 1 Environment variable Condition2 Value of the eperimental parameter Condition defining group 2 Environment variable Q cutoff Q value cutoff GeneSpring GX R integration package 37 38 Get the libraries that are needed library graphics library stats utils siggenes library library library GeneSpring memory limit size 2000 Get environment variables Param Sys getenv c ExperimentParameter Conditionl Condition2 Q cutoff ExperimentParameter lt as character Param 1 Conditionl as character Param 2 Condition2 as character Param 3 cutoff lt if as numeric Param 4 lt 0 0 else if as numeric Param 4 gt 1 1 else as numeric Param 4 Get the GeneSprint Experiment gs int GSload exp the parentheses and find it in
26. Graphical User Interface of in GeneSpring GX Except for command line version of R Rcmd exe it is also possible to use the R Graphical User Interface Rgui exe to run R scripts The GeneSpring GX R integration package can also run any of the scripts in this interface and the integration package includes a special bat file that can be used to run any of the scripts in the GUI of R Running scripts in the GUI of R makes it possible to plot graphs and use any of the other GUI objects in R like the graphical file browser fileBrowser for example 3 4 1 3 GS exec GUlI bat The bat file that will execute an R script in the GUI of R is called GS exec GUI bat and the contents of the file with some short explanations is given below Gecho off copy 1 Rprofile copy output txt shift Loop IF GOTO Continue set 1 2 shift shift GOTO Loop Continue cat exe GS R in txt rw2010 bin Rgui exe no save no restore cat exe GS R out txt del GS R in txt GS R out txt copy output txt Rprofile Most of the lines are similar to those in the non GUI execution bat file explained above with the exception of the lines copy 1 Rprofile copy output txt and rw2010 bin Rgui exe no save no restore GeneSpring GX integration package 21 22 3 4 2 3 4 3 Since it is not possible to execute a script with the GUI interface by providing the script on the command line using the STDIN redirect
27. HEE HEHE EE EEEE EEEE EEE EEEE EEEE EEEE EEEE EEEE EEE EE EEE EEE EEE EE EEEE gapFilter R This script implements the BioConductor gapFilter genefilter The gapFilter looks for genes that might usefully discriminate between two groups possibly unknown at the time of filtering To do this we look for a gap in the ordered expression values The gap must come in the central portion we exclude jumps in the initial Prop values or the final Prop values Alternatively if the IQR for the gene is large that will also pass our test and the gene will be selected The script expects RAW data from an Interpretation to be sent from GeneSpring Negative values and NA values will be removed before the filtering Version 1 0 0 Author Thon de Boer Agilent Technologies SiliconGenetics Redwood city CA Last Update 28 May 2004 Environment variable Gap The size of the gap required to pass the test Environment variable IQR The size of the Inter Quartile Range IQR required to pass the test Environment variable Prop The proportion or number of samples to exclude at eitherend dE dE dE dE dE dE db de d de dE dE db de dE dE db dE db dE db FEE AE HE AEE AE E FE AE FEAE AE E AE FE E AE E AE E AE E AE FE E FE AE FE AE AE AE FEAE FEAE FEE AEE EEE H Prepare GeneSpring environment library graphics library stats library utils library GeneS
28. S exec R GUI bat boxplot r cv filter Fi DifferentialExpression LPE R gapFilter R power anova R qvalue R Z score H cat exe folder rw2010 OPTIONAL Moving the files is only required if you have changed your GeneSpring GX data directory 3 3 2 UNIX MacOSX 3 3 2 1 Change the path to the GS exec R sh script REQUIRED By default GeneSpring GX is installed in the users home directory SiliconGenetics GeneSpring GX and the External Program definitions have to be changed to reflect the location of the scripts Edit the External Program definitions and change the path to the program GS exec R sh From the main GeneSpring GX window open the External Programs folder and select the external program definition that you would like to edit by selecting Inspect from the RIGHT CLICK pop up menu Full Development MG U74 All ver2 Genes all genes SE File Edit View Experiments Colorbar Filtering Tools Annotations Window Help 2 Filter CV B 100 gt E Run He E Rename Delete G Open Interpretation in Q Q Plots g Qvalue g Qvalue with plots RMA normalize CEL files Y axis RATIO AllRef De RMA normalize CEL files interactive Colored by Time 1 hours Ho A Save Classification To File List all genes 36767 Save Experiment To File E save Gene List To File Cauna Nana l int iiith hiimbhara Ta m m e o
29. S qval i n file filename append T Original code by Storey stops here GeneSpring environment library GeneSpring Get the Parameters and do some sanity checking Param Sys getenv cutoff bootstrap robust Ocutoff lt if Param 1 gt 1 1 else if Param 1 0 0 else as numeric Param 1 bootstrap if Param 2 no bootstrap else smoother robust if Param 3 no TRUE else FALSE Getting the data g list GSload genelist qobj lt qvalue g list 2 fdr level Qcutoff pi0 meth bootstrap robust robust genelist data frame genes g list qobj significant 1 qvalue qobj qvalues qobj significant Save the gene list GSsave genelist genelist the plots if we are in an interactive session if interactive qplot qobj 5 8 Z score normalization This script will normalize an experiment using the Z score method HHH REE HE E IEE EAE E IE EAE AE AE E EAE AE AE EE AE AE E FE AE AE AE EAE AE AE E FE AE AE AE AE E E AE FE FE E AE AE AEE E AE AE EE E AE AE E E AE AE AE AE E AE AEE E AE AE AE FE E AE AE AE E AE E AE AE EAE AE AE AEE EE EEE EE Z score R Program that reads a GeneSpring Experiment and returns a new GeneSpring experiment with this data again normalized using the Z score method Version 2 0 0 Author Thon de Boer Agilent Technologies SiliconGenetics Redwood city CA Last Update 29 Apr 2005 FEE AE HE AEE AE E AE
30. alues gs int a nor lt na exclude gs int a nor Add the parentheses and find it in the list of parameters quit if we don t No point doing anything if we don t know which parameter to use i lt charmatch ExperimentParameter row names gs int expparam if is na i qi Preprocess the data Normalize with dChip algorithm and LOWESS set the threshold to a very small value in case we get pre normalized data expr lt preprocess gs int a nor data type dChip LOWESS TRUE threshold 1E 3 Find the data defining the two groups x lt expr gs int expparam i Conditionl y lt expr gs int expparam i Condition2 Calcualte the baseline error distribution in 100 bins x var baseOlig error x q 0 01 y var baseOlig error y q 0 01 Do the lpe val lt lpe x y x var y var probe set name gs int genenames Adjust the P value Could choose from c BH BY resamp none no I it is no we ll do it anyway but send back the Raw p value if is na grep MulTestingCorrection c BH BY resamp 1 lpe MTC lt my fdr adjust lpe val adjp BH else if MulTestingCorrection resamp lpe MTC fdr adjust lpe val adjp resamp target fdr PvalueCutoff else lpe MTC my fdr adjust lpe val adjp MulTestingCorrection Create the gene lists if MulTestingCorrection resamp geneindex lt lpe val z stats lpe MTC 1 z critical genelist lt d
31. and GeneSpring GX for Windows and Unix platforms For more information on R visit the web site http cran us r project org What is new in this release This is the second release of the GeneSpring GX R Integration package and a number of new features and bug fixes are part of this release 1 Upgrade R to version 2 1 0 The version of R that is included in this release is version 2 1 0 released 18 Apr 2005 This contains numerous changes and bug fixes For a complete list of the new features in R 2 1 0 consult the R web site 2 Added box plot script Added new example script to create a simple Box Plot in R 3 Removed RMA scripts Since GeneSpring GX 7 1 has native support for RMA and GC RMA analysis of CEL files there is no need for a R version of the RMA scripts and therefore have been removed from this version 4 All of the scripts were updated to work with GeneSpring GX R library version 2 0 0 BioConductor 1 6 contains a new version of the GeneSpring GX library version 2 0 0 and the example scripts were changed to work with GeneSpring GX library version 2 0 0 GeneSpring library version 1 0 x used the GeneSpring GX systematic name as row names for many of the expression matrices and certain characters that are valid characters in systematic names Namely are not valid characters for Matrix row names leading to some conversion issues GeneSpring GX library 2 0 0 stores the gene names in a special slot allowing for
32. as a space in the name Like windows Program Files folder you need to quote the directory name with double quotes For example if is installed in the default directory on Windows the path to the Rterm exe program is C Program Files R rw2010 bin Rterm exe no restore slave lt R SCRIPTS gt Rterm output txt Note the quotes around Program Files Any folder that has a space in the name needs to be quoted like this Also edit the GS exec R GUlI bat script to reflect the different location of the R program Edit the line rw2010 bin Rgui exe no save no restore to something like YOUR PATH HERE Rgui exe no save no restore Remember to quote any folders with spaces in the name Note that the GS exec GUl bat script uses a different R program i e Rgui exe and there are no input or output files This program will receive the R script in a different manner as will be explained in other sections 3 3 1 2 Non default data directory location OPTIONAL If the preference setting for the location of the GeneSpring GX data directory has been changed in the past you will need to move the bat files the R scripts and the rw2010 folder If you chose to install the R program to the original data directory in order for the scripts to run successfully If you have changed your data directory settings move the following files to the original data directory GS exec HR bat GeneSpring GX R integration package G
33. ata frame genes row names lpe val geneindex Z stats lpe val geneindex z stats else Get the column name column lt paste p adj adjp ifelse is na grep MulTestingCorrection c BH BY 1 rawp MulTestingCorrection Create the GeneList geneindex lt lpe MTC column lt PvalueCutoff genelist lt data frame genes row names lpe MTC geneindex FDR lpe MTC geneindex column Write the resulting Gene List GSsave genelist genelist 5 5 BDifferentialExpression SAM he program will calculate the d value for differential expression between the two groups using the SAM Significance Analysis of Microarrays method implemented in as described by Tusher V G Tibshirani R and Chu G 2001 Significance analysis of microarrays applied to the ionizing radiation response PNAS 98 5116 5121 NOTE A LICENSE IS REQUIRED FOR COMMERCIAL USE OF SAM Non commercial use is permitted For more information on how to obtain a license for commercial use of SAM see http otl stanford edu industry resources sam html HE HEE HEE BEEBE HEH BE ERE ERE PERRET HE RAE ERE HEH SAM R Program that will accept an experiment from GS and return a genelist with d values The program will calculate the d value for differential expression between the two groups using the SAM method as described by Tusher V G Tibshirani R and Chu
34. ction of GeneSpring GX should contain a number of extra items as shown in the figure below Full Development Perou custom array Genes all genes DER File Edit View Experiments Colorbar Filtering Tools Annotations Window Help Expression Profiles 1007 EH External Programs 1 i Calculate Power on Interpretation DifferentialExpression LPE DifferentialExpression SAM DifferentialExpression SAM w plots F Filter cv 09 gapFilter HB Load Classification From File HE Load Experiment From File M Load Gene List From File He Load Gene List With Numbers From File He Load Gene Tree From File H Qvalue Qvalue with plots H RMA normalize CEL files RMA normalize CEL files interactive Save Classification To File Save Experiment To File y H Save Gene List To File 0 01 i 2 y _ Ann i Normal CR1 2276 b Ductal CR1 2154 tt Fibroadenoma CRB1 3200 bt Lobular CRB1 nod an normal Normal diseased Ducti diseased Lobular Send Experimentto Pathwaysssist Y axis Perou Breast Cancer Study Default Interpretation send Gene PathwayAssist Colored Normal CR1 2276 b normal Normal send to SRS Gene List all genes 9539 Z score normalization BE Rankmarke 4 Normalized Intensity log scale Expression Snow al Data Magnification 1 There are 8 or 11 Depending on wheth
35. ctor A Calculate Powe 4f Display express Expand network A on Colored P Disease tissue Sub Gene List all genes 12585 ry Expression A Load Classifica H Load Experimer In order to run the External Program you simply click the External Program s name and the script will be run If you have defined any Arguments a new dialog box will pop up asking you to fill in the values for the Arguments Since we have defined two arguments for the CV filter program a dialog box appear GeneSpring integration package User Parameters for Filter DER Enter values for the following parameters CV cutoff hoo Matching_conditions 1 OK You can change the values if the default values are not appropriate and press lt OK gt to start the actual script You will not see any activity except for the hourglass that indicates the filter is running If the filter is complete a results window will appear and you will be able to save the results in GeneSpring GX 5 New Gene List 11 287 genes Filter on CV EC Gene Lists 100 g atRab geranylgeranyltransferase alpha subunit C3 Disease type study 1000 atmitogen activated protein kinase 3 1001 attyrosine kinase with immunoglobulin and epidermal growth fe Pathways 1002 f atH sapiens mRNA for cytochrome P 450 HA Prostate cancer study 1003 s atBurkitt lymphoma receptor 1 GTP binding protein 1
36. d 4 robust Parameter will determine if the robust method for small P values will be used Set to yes to use the robust method set to no for using the default method Output 1 GeneList containing the genes that pass the filter with the Q value as the associated value GeneSpring GX R integration package 3 2 2 6 8 Q value with plots from 1 way ANOVA results WINDOWS ONLY This script performs the exact same calculation as the previous script but this script will actually show four plots in the R Graphical User Interface as shown below 7 RGui Graphics Device 2 ACTIVE R File History Resize Windows 0 615 0 4 0 6 0 8 1 0 significant tests u E wo a co 2 E o co a gt v 0 00 0 02 0 04 0 06 0 08 0 10 0 2000 6000 q value cut off significant tests The plots can be used to determine an appropriate Q value cutoff and will give information about the number of expected false positives 3 2 2 6 9 Z score Normalization This script creates an experiment with Z score normalization using all the samples of the selected experiment Z score ExprVal Mean Input 1 Experiment All samples in the Experiment will be sent to the external program Knobs There are no parameters to be set for this script Output 1 Experiment A new experiment is presented to the user with all the values normalized with the Z score method centered around 10
37. e 4 FDR The False Discovery Rate as a fraction 1 1 The FDR will be used to determine the d value cutoff values of what is to be considered a significantly differentially expressed gene Output 1 GeneList Those genes that are differentially expressed between the two groups and result in a FDR as set by the FDR knob The associated value for the gene list will be the d value as calculated by the SAM method 3 2 2 6 5 Filter on CV This script will filter the chosen Interpretation on the Coefficient of Variation CV 100 Standard Deviation Mean The CV values can be calculated on either the Raw data Default or the Normalized data The script produces a Gene List of those genes that show a CV value less than the CV cutoff parameter in a specified number of Conditions Input 1 Experiment Interpretation 2 The starting gene list Knobs 1 CV cutoff The minimum Coefficient of Variation in percent 2 Matching conditions The minimum number of Conditions that have to show a CV value less than CV cutoff The Matching conditions parameter is either a whole number or a fraction When a whole number it indicates the absolute number of conditions that need to pass when a fraction it indicates that fraction of conditions that need to pass For example a setting of Matching conditions of 4 indicates 4 conditions need to pass while 0 5 indicates that half of the conditions need to pass 3 Data to use T
38. e output FROM the R script is for the script The INPUT tab defines what the R script will use as input and in the case of the CV filter that would be the Experiment Condition Statistics Click on the INPUTS tab if it is not already the default tab selected and click on the lt ADD INPUT gt button A new dialog box is displayed Choose Experiment Condition Statistics from the left part of the dialog to choose this as the input for the R script When the Experiment Condition Statistics is selected the right part of the dialog box become active and here you can choose what data you would like to send to the R script Since the CV filter will be using the Average of the Normalized value and the Standard Deviation of the Normalized values select the checkbox next to the Average and Standard Deviation labels under the first column which is called Normalized Values and press OK 5 Choose Type of Input to External Program Choose Input Type Values Included in Input Gene List Normalized Values Control Values Raw Values C Gene List With Numbers Average Average Average C Gene Name Experiment Data Minimum Minimum Minimum C Experiment Data with Confidence Maximum Maximum Maximum C Standard Error Standard Error Standard Error C Tree Standard Deviation Standard Deviation C Genome Experiment Condition Statistics Number of Samples Number of Samples Number of Samples OK Cance
39. e used for the R programs that you may write and the icon file is called Rlogo16 gif GeneSpring GX R integration package 23 24 3 4 4 1 And you may choose this icon by pressing the Browse button and selecting the file which is located in the GeneSpring GX data directory Program Executable GeneSpring GX supports three different types of External Programs The proper External Program or command line tool A JAVA class or an HTTP call In the case of executing an R script we will leave the default setting External Program Command line In the command line text box we will type the following GS exec R bat cv filter The first part is the R executing Batch program and the second part is the actual R script The second part could be any R script that you write and in this case we have used the example R script cv filter NOTE On UNIX systems the complete path to the GS exec R sh file needs to be given If GeneSpring GX was installed in the default location the command line should read opt SiliconGenetics GeneSpring data GS exec H sh cv filter R Also note that since the R scripts are stored in the same directory as the script it is not needed to provide the full path to the script If you choose to store your scripts in a different location than the R execution scripts adjust the paths accordingly Input Output Delimiters Arguments Tabs The next section defines what the input FOR the R script and th
40. ed by Jain et al Bioinformatics Vol 19 1945 1951 2003 If the BH or BY multiple testing corrections are applied the FDR is returned as the associated value If the resampling method is used the associated value represents the Z score for the gene Version 2 0 0 Author Thon de Boer Agilent Technologies Redwood city CA Last Update 17 Feb 2005 Environment variable ExperimentParameter Name of the experimental parameter that define the groups Environment variable Conditionl Value of the eperimental parameter Condition defining group 1 Environment variable Condition2 Value of the eperimental parameter Condition defining group 2 Environment variable PvalueCutoff P value cutoff value Only genes with CORRECTED P values below this value are returned Environment variable MulTestingCorrection Type of mutliple testing correction to use db ck db cb db db db b GR dE dE db THEETSERSEREREEEERETRETESARAHTATEAHEHEERHATHRSAEHAARETETERETEEERETEEAETETETERETEEEREEREEETETETHEEEETESERTEEETIEBHE Get the libraries that are needed library graphics library stats library utils Get the library library LPE GeneSpring environment library GeneSpring memory limit size 2000 I had to re implement the fdr adjust function for the BH or BY MTC since the version contains errors it seems It was ordering the matrix but it lost the rownames
41. er you installed on Windows or on UNIX new External Programs defined in the folder External Programs that are indicated with a R icon q y 1 Boxplot WINDOWS ONLY 2 Calculate Power on Interpretation DifferentialExpression LPE 4 DifferentialExpression SAM 5 DifferentialExpression SAM w plots WINDOWS ONLY 6 Filter CV GeneSpring GX R integration package sy gapFilter sy Qvalue y 9 Qvalue with plots WINDOWS ONLY 10 Z score normalization Each of the external programs is implemented as an R external program via the EPI Although the external programs can be run from the External Programs section we have also provided a number of scripts that implement the R scripts that will give the user a little more control over how the R scripts should be run and will also show the integration of the R external programs into the scripting environment of GeneSpring GX 3 2 2 6 New Scripts The installer has also created a new folder in the Scripts folder called R scripts that contains eight new scripts amp Full Development Perou custom array Genes all genes File Edit View Experiments Colorbar Filtering Tools Annotations Window Help EC Experiments 100 Gene Trees Condition Trees Classifications Pathways Array Layouts Expression Profiles External Programs Bookmarks EH Scripts J Basic Scripts External Programs Hod R scripts B Calculate power HB Differentia
42. fined for this script Output This script does not produce any output that can be stored in GeneSpring GX lt Graphics Device 2 ACTIVE Ele History Resize Toxicogenomics Test Set 1651 genes o o 8 3 2 2 6 2 Calculate power This script calculates the Power for each gene based on the variance within and between the groups conditions of the Interpretation Input 1 Experiment condition The input of this script is an Experiment Interpretation containing at least two conditions and each condition should have at least 2 replicates to be able to calculate the variation within the group Knobs 10 GeneSpring integration package 1 SignificanceRequired The power will be calculated using this value as the required significance Output 1 Gene List with the power as the associated value 3 2 2 6 Differential Expression using LPE This script will filter the genes with an adjusted P value less than the cutoff for differential expression between the two conditions using the Local Pooled Error Model as described by Jain et al Bioinformatics Vol 19 1945 1951 2003 The local pooled error test attempts to reduce dependence on the within gene estimates in tests for differential expression by pooling error estimates within regions of similar intensity Note that with the large number of genes there will be genes with low within gene error estimates by cha
43. h T las 2 main paste i expName i ngenes genes sep 5 2 Filter on CV This script will take an Experiment Interpretation as input and produce a Gene List as output of all the genes that pass the CV_cutoff and Matching_conditions settings FERRER HE RES EEEE HERE ES HEE ESHER HAE RHEE ERE HAE EEEE EEE EEEE EEEE EEEE EEE EEEE EEEE EEEE EEEE EEE EEEE EE HF k db b Hb cb Hb dk od cv filter R Program that reads a GeneSpring Experiment Interpretation and creates a Gene List of those genes that fall within the parameters for the Coefficient of Variation Version 2 0 0 Author Thon de Boer Agilent Technologies Redwood city CA Last Update 17 Feb 2004 Environment variable CV cutoff Value for the cutoff for the Coeficient of Variation Environment variable Matching conditions Number of conditons that need to pass Can be ratio less than 1 or absolute number 1 Environment variable Data to use Determines whether the CV should be calculated on the Raw R or r or the Normalized data N or n JHREHEHEEHHEHEREEHEHEHEEHEHEHEREEHEHEREEREHEEHEEHEHEHEHEREREHEEHEREREEHEEHEHEEHEHEEHEHEEHEREREREHEEHEHEEHEREREREHEEHEREHEREHEHEEEHEEHEHEEREHEREHEE REE RE Prepare GeneSpring environment library graphics library stats library utils GeneSpring integration package library GeneSpring Read the Experiment Interpretation gs int GSload
44. he type of data to use Possible values are Raw or Normalized R or N can also be used Default Raw Output 1 GeneList Those genes that pass the filter cutoff and the number of matching conditions 3 2 2 6 6 Gap Filter This script implements the BioConductor filter gapFilter The gapFilter looks for genes that might usefully discriminate between two conditions possibly unknown at the time of filtering To do this we look for a gap in the ordered expression values The gap must come in the central portion we exclude jumps in the initial Prop values or the final Prop values Alternatively if the Inter Quartile Range IQR for the gene is large that will also pass our test and the gene will be selected Input 1 Experiment Interpretation The NORMALIZED data for each of the genes for each Condition will be used GeneSpring integration package 13 14 Knobs 1 Gap The minimal Gap required between the high and low values for the gene to pass the filter 2 IQR Inter Quartile Range When a gene shows a IQR of at least this value it will also pass the filter 3 Prop The proportion of values that should be excluded from Gap and IQR calculations If less than 1 the value is taken as a ratio if 1 or higher the number represents the number of conditions on each side of the sorted values to be disregarded Output 1 GeneList Those genes that pass one or both of the filters
45. i e a z a S69 Agilent GeneSpring GX R integration package Installation and Application Guide E Agilent Technologies GeneSpring GX integration package 1 Table of Contents L Table OP CODnItDIsa reser teo neenon 3 2 Introduction and summary 3 2 3 2 1 1 e ON 4 2 2 What is new an this release eterne oret boe 4 2 3 X er 4 3 The GeneSpring GX External Program eene 5 3 1 How the EPI Works 5 3 2 5 3 2 1 Upgrading from a previous version of integration 6 3 2 2 Installation procedure 2 9 2 n Reiete 6 3 3 Post installation nne 16 3 3 1 16 3 3 2 5 X qr 17 34 Creating External Programs using 19 3 4 1 WINDOWS 5_ _ 19 3 4 2 UNIX Mac OS X GS 8 22 3 4 3 Create the R 22 3 4 4 Create the External Program definition in GeneSpring
46. ine If the script had two arguments called CV cutoff and Matching conditions and the values set to be 100 and 3 this will be converted into two environment variable of the same name with the value 100 and 3 cat exe GS R in txt This line re directs the output from GeneSpring GX which could be the experiment data or the selected gene list for instance to a file called GS in txt using the program cat exe which is installed by the GeneSpring GX R Integration package installer This GS in txt file is a temporary file that will be used by the R script later We re direct the data from GeneSpring GX into a file in the GeneSpring GX data directory since it is somewhat easier to deal with in R but it is not absolutely necessary We could elect to read the STDIN stream directly in R The working directory for any of the external programs will be the GeneSpring GX data directory which on WINDOWS System is C Program Files SiliconGenetics GeneSpring data If the location of the data directory has been GeneSpring GX R integration package changed to a different location the working directory WILL STILL BE THE OLD DIRECTORY rw2010 bin Rterm exe no restore slave lt R SCRIPT gt Rterm output txt This line will execute the R program and reads in the R script The program we are using from the package is Rterm exe which is the program that can execute R scripts without the GUI interface Later we will describe an e
47. ition2 The value of the Experimental Parameter condition that defines group 2 NOTE The ExperimentParameter and Condition values are case sensitive so ensure that you use the correct case for the parameter name and values 4 PvalueCutoff The P value cutoff value Only genes with an adjusted P value of this value or less will be returned b MulTestingCorrection Name of the multiple testing correction method to be used to adjust the P value Possible values for the Multiple Testing corrections are For false discovery rate FDR BH Benjamini amp Hochberg 1995 Default BY Benjamini amp Yekutieli 2001 none or no No Multiple Testing Correction GeneSpring GX integration package 11 12 Output 1 GeneList Those genes that are differentially expressed between the two groups and have a adjusted P value below the cutoff PvalueCutoff 3 2 2 6 4 X Differential Expression using SAM The program will calculate the d value for differential expression between the two groups using the SAM Significance Analysis of Microarrays method implemented in R as described by Tusher V G Tibshirani R and Chu G 2001 Significance analysis of microarrays applied to the ionizing radiation response PNAS 98 5116 5121 NOTE A LICENSE IS REQUIRED FOR COMMERCIAL USE OF SAM Non commercial use is permitted For more information on how to obtain a license for commercial use of SAM see http otl stanf
48. l Expression using LPE HB Differential Expression using SAM i Filter on cv 0 01 L LS Gap Filter Normal CR1 2276 bd Ductal CR1 2157 bt Lobular CR1 2174 bd L amp aq value from 1 way ANOVA results Normal diseased Ductal 1 1 diseased Lobular re Q value with plots from 1 way ANOVA results Y axis Perou Breast Cancer Study Default Interpretation HB Normalize CEL files Colored by Normal CR1 2276 t normal Normal HE Normalize CEL files Interactive Gene List all genes 9539 S Zscore Normalization EHA Unleann Pinte 24 gt Pa Show All Data Normalized Intensity log scale Expression N 1 f 7 1 Magnification 1 The scripts are called gt plot Calculate power Differential Expression using LPE iN Differential Expression using SAM Filter on CV Gap Filter value from 1 way ANOVA results value with plots from 1 way ANOVA results 9 Z score Normalization The scripts contain extensive notes explaining how they work A short overview of the purpose of the scripts is given below GeneSpring GX integration package 9 3 2 2 6 1 This script will create a box plot in a graphical R window Input 1 Experiment Interpretation The interpretation for which the box plot needs to be created Knobs No Knobs are de
49. l Show Example GeneSpring integration package Next select the OUTPUTS tab and click the ADD OUPUT gt button and a new dialog is displayed In this case we will select the radio button next to the label Gene List With Numbers since the result of the filter will be a Gene List where the associated values will contain the number of conditions that passed the filter You can give the column of Associated Values a name by typing something in the text box at the bottom labeled Description of Numbers In this case we have used Passed as the label for the column of associated values Click OK to accept the settings 5 Choose Type of Output from External Pro DEAR Choose Output Type C Gene List Gene List With Numbers C Gene Name C Experiment Data C Experiment Data with Confidence C Classification C Tree C Genome C Temporary Genome Label for Gene List With Numbers Description of Numbers Passed OK Cancel Show Example Select the tabs labeled Arguments In this tab we will define the arguments that will be passed to the script that will define what criteria the filter should use in determining if a gene passes or fails In the example we are using here there will be two criteria The first filter criterion is the value of the minimum Coefficient of Variation that we will tolerate for a given gene and the second criterion is the minimum number of conditions that should have this CV value Add
50. ms into a common platform like GeneSpring GX The GeneSpring GX External Program Interface EPI will allow a user to integrate any program or algorithm in a seamless way giving the user the flexibility to perform whatever analysis they would like in the familiar environment without having to manually save the expression data outside of GeneSpring GX and perform the analysis and re import the data into GeneSpring GX With the EPI this can all be done in such a way that the end user will never have to even see that an analysis is performed outside of GeneSpring GX GeneSpring GX integration package 3 2 2 2 3 Many scientists have chosen to perform the specialized analyses in the Language R in special R scripts and the GeneSpring GX R integration package was developed to facilitate the incorporation of those R scripts into the GeneSpring GX environment The integration of the R language into GeneSpring GX is only one of the examples showing the versatility of GeneSpring GX to interact with other analysis tools and we hope that you will find this information useful in your future expression analysis research R R is a statistical computing language which is freely available It is rapidly gaining traction within the Bioinformatics world and especially in the Gene Expression analysis field since it has a high statistical analysis need The R program can be obtained for many platforms and this document will cover the integration of R
51. n your machine download the file from the GeneSpring GX Extra section of our website http www agilent com chem genespring or use the links provided in the previous section There are a number of versions of the installer depending on your platform WINDOWS UNIX MacOSX and whether you would like to include R or not Open GeneSpring GX and once the window for the Genome is opened go to the location where you saved the ZIP file and select the file and drag the file onto the open window of GeneSpring GX In this case it does not matter which Genome you have opened it will work with any Genome Alternatively use the menu item File Open GeneSpring GX Zip to open the ZIP file Once you dropped or opened the file in GeneSpring GX a dialog box will be presented as shown below amp Install R integration package This will install the R integration package containing R Version 2 1 0 BioConductor Version 1 6 and a number of example scripts for integrating R scripts into GeneSpring See the Document GeneSpring R Integration package pdf in GenSpring data directory for more information GeneSpring will quit when this operation 1s finished Do you wish to proceed wo The dialog box shows a short description of the Installer and it will inform the user that GeneSpring GX will quit after the installation is complete Press No if you are not ready to quit GeneSpring GX yet press Yes if you would like to proceed
52. name of the GeneSpring GX package is GeneSpring Note the capitalization of the word GeneSpring At this point R version 1 9 1 the capitalization is not strictly enforced but in the future it may be so take care to refer to the package always in the correct capitalization Making the library active can be performed by using the command library GeneSpring After the package has been installed A list of functions and class definitions provided by the GeneSpring package is given below and are described in more detail in the next section Functon name Description GSint Create GeneSpring Experiment Interpretation object class GSint GSload int Load GeneSpring Experiment Interpretation from file and create Interpretation object class GSint GSload intBC Load GeneSpring Experiment Interpretation from file and create BioConductor expression object class exprSet GSIoad exp Load GeneSpring Experiment from file and create Interpretation object class GSint GSload expBC Load GeneSpring Experiment from file and create BioConductor expression object class exprSet GSload genelist Load GeneSpring gene list from file GSsave exp Save GeneSpring Experiment to file from Interpretation object class GSint GSsave genelist Save GeneSpring gene list to file GSint2BC Convert GeneSpring Interpretation object GeneSpring GX R integration package 33 5 1 clas
53. nce so that some signal to noise ratios will be large regardless of mean expression intensities and fold change The local pooled error attempts to avert this by combining within gene error estimates with those of genes with similar expression intensity The script produces a Gene List of those genes that show differential expression between two groups indicated as the input parameters knobs The adjusted P value will be provided as the associated number for the genes Input 1 Experiment Interpretation Since the script will perform a dChip and Lowess normalization it is not required that the data is prenormalized If normalized data is used the dChip and Lowess normalization will be performed on pre normalized data which should not have an adverse effect on the outcome but could result in different gene lists NOTE THE LPE ALGORITHM CANNOT DEAL WITH MISSING VALUES AND GENES WITH MISSING VALUES WILL NOT BE USED IN THE CALCULATION Knobs 1 ExperimentParameter The name of the Experimental Parameter that contains values that can be used to distinguish the two groups NOTE The parameter name should not contain spaces If the parameter DOES contain spaces only use the first word of the parameter name For instance in the parameter is called Disease type use Disease only Use enough characters to uniquely identify the parameter 2 Condition The value of the Experimental Parameter condition that defines group 1 3 Cond
54. o other changes are needed and we have provided an R execution BAT file to execute scripts that require a Graphical Interface The BAT file is called GS exec R GUl bat Use this bat file in stead of the standard GS exec file and the script will be executed in the Graphical environment of R GeneSpring GX integration package 21 4 1 The provided External Program definition Q value with plots is an example of an R script that is using the graphical environment of R in its execution The definition of the External Program is given below amp Edit External Program Details Program Executable External Program Java Class C HTTP Command Line GS exec R GULbat qvalue R Browse Inputs outputs Delimiters Arguments Input from GeneSpring to External Program Add Input Gene List With Numbers Debug Input Save Help Note that the only difference with the definition for the External program value is the fact that the GS exec file is used to execute the R script Also note that the R script used in both external programs is identical The PLOT command in the script is actually present in the NON GUI version of the execution but the plot is simply not shown 4 The GeneSpring Package The GeneSpring R package is part of the standard BioConductor offering and can be installed with the standard mechanisms available in R and Bio
55. on of null p values qvalues a vector of the estimated q values the main quantity of interest pvalues a vector of the original p values significant if fdr level is specified and indicator of whether the q value fell below fdr level taking all such q values to be significant controls FDR at level fdr level GeneSpring GX R integration package 39 This is just some pre processing if min p 0 max p 1 print ERROR p values not in valid range return 0 if length lambda gt 1 amp amp length lambda lt 4 print ERROR If length of lambda greater than 1 you need at least 4 values return 0 m lt length These next few functions are the various ways to estimate 10 if length lambda 1 pid lt mean p gt lambda 1 lambda pid min pi0 1 else pid lt rep 0 length lambda for i in 1l length lambda 10 1 lt mean p gt lambda i 1 1 if pi0 meth smoother spi0 lt smooth spline lambda pi0 df 3 pid lt predict spi0 x max lambda y pid lt min pi0 1 if pi0 meth bootstrap minpiO lt min pi0 mse rep 0 length lambda pi0 boot lt rep 0 length lambda for i in 1 100 p boot lt sample p size m replace T for i in 1 length lambda pi0 boot i lt mean p boot lambda i 1 lambda i mse lt mse pi0 boot minpi0 2 pid lt min 0 mse piO
56. ord edu industry resources sam html The script produces a Gene List of those genes that show differential expression between two groups indicated as the input parameters knobs The d value will be provided as the associated number for the genes The script will launch the R program and will also produce a plot in the R interface that can be used to determine a good FDR and d value cutoff values When the R program is closed the resulting gene list can be saved in GeneSpring GX R Graphics Device 2 ACTIVE Delta vs Delta vs Significant Genes R Console Input 1 Experiment Interpretation The normalized experiment data will be used Knobs 1 ExperimentParameter The name of the Experimental Parameter that contains values that can be used to distinguish the two groups NOTE The parameter name should not contain spaces If the parameter DOES contain spaces only use the first word of the parameter name For instance in the parameter is called Disease type use Disease only Use enough characters to uniquely identify the parameter 2 Condition1 The value of the Experimental Parameter condition that defines group 1 3 Condition2 The value of the Experimental Parameter condition that defines group 2 NOTE The ExperimentParameter and Condition values are case sensitive so ensure that you use the correct case for the parameter name and values GeneSpring GX R integration packag
57. ors Type contributors for more information and citation on how to cite or R packages in publications Type demo for some demos help for on line help or help start for a HTML browser interface to help q to quit gt utils setRepositories gt utils menuInstallPkgs Error in install packages NULL libPaths 1 dependencies TRUE type typ no packages were specified gt A gt R 2 1 0 A Language and Environment B A list of available CRAN mirrors will appear Choose the CRAN mirror that is closest to your physical location to ensure the fastest response and press OK GeneSpring GX R integration package CRAN mirror France Lyon France Paris Germany Berlin Germany Koeln Germany Mainz Germany Muenchen Hungary Italy Arezzo Italy Ferrara Japan Aizu Switzerland Zuerich Switzerland Bern 1 Switzerland Bern 2 Turkey Taiwan UK Bristol UK A list of available packages will appear Select the package GeneSpring from the list and press OK GeneSpring GX integration package 31 genefilter Geneland GeneMeta GeneNT geneplotter GeneR geneRecommender GeneSprin genetics GeneT raffic GeneTS GeneTS GenKem GLMMGibbs glramML The R program will download the package from the BioConductor website and install the library files in the correct location After the installation is complete Press y to allow R
58. pring library Biobase library genefilter Read the Experiment Interpretation and Create BioConductor exprSet gs int GSload int eSet GSint2BC gs int what raw Get the Parameters and do some sanity checking Param Sys getenv c Gap IQR Prop Turn parameter into variables Gap lt if Param 1 lt 0 0 else as numeric Param 1 lt if Param 2 lt 0 0 else as numeric Param 2 Prop if Param 3 1 1 else as numeric Param 3 Create the filter function Gfilter gapFilter Gap IQR Prop ffun filterfun Gfilter Filter the genes filteredEset genefilter eSet ffun GeneSpring GX integration package 35 36 Create the gene list genelist lt names filteredEset filteredEset Write the resulting Gene List GSsave genelist genelist 5 4 DifferentialExpression LPE The program will calculate the P value for differential expression between the two groups using the Local Pooled Error Model as described by Jain et al Bioinformatics Vol 19 1945 JHREHEHEEHHEHEREEHEREHEEHEHERHEHEEHEEHEREEHEHEEHEEHEHEHEHEEHEHEEHEREREEHEEHEHEEHEREEREHEEHEREHEREHEEREHEEHEHEREEHEEHEREREREHEHEEHEHEEHEREEREREEHEE REE RE DifferentialExpression LPE R Program that will accept an experiment from GS and return a genelist with P values The program will calculate the P value for differential expression between the two groups using the Local Pooled Error Model as describ
59. pring GX Library from the CRAN BioConductor site by using the built in download packages functionality in R See Chapter 4 for more information 3 Download one and only one of the GeneSpring GX R Integration package ZIP file from our web site listed below The installation of GeneSpring R integration package via the drag and drop interface of the ZIP file is only compatible with version 6 1 of GeneSpring and later For information about upgrading to version 6 1 or later released Nov 2003 see the Agilent website at http www agilent com chem genespring or chose Update GeneSpring from the Help menu GeneSpring GX integration package 5 For WINDOWS including R program 2 1 0 and BioConductor GeneSpring library http www chem agilent com cag bsp SiG Downloads zip gs r win for WINDOWS NOT including R program any library http www chem agilent com cag bsp SiG Downloads zip gs r win Zip OR for UNIX NOT including program or any library http www chem agilent com cag bsp SiG Downloads zip gs r uni 4 Drag the ZIP file onto any open window of GeneSpring GX or use Menu File gt GeneSpring Zip 5 Change the path for the scripts and or R program See section 3 3 6 Change the permission on the execution scripts UNIX only See section 3 3 2 7 PRESTO You are done It s that simple to get started with R in GeneSpring GX If you need more help in the installa
60. s GSint to a BioConductor expression object exprSet BC2GSint Convert a BioConductor expression object exprSet to a GeneSpring Interpretation object class GSint More information about the functions can also be obtained by using the help or function in R 5 Appendix A Complete R examples This section contains the example scripts that are provided with the GeneSpring GX R integration package Read the comments for more information and on hints for writing your own scripts Box plot This script will take an Experiment Interpretation as create a Box plot in a graphic window No object is returned AAEE RAEES EEEE EEEE EEEE EE HE HEA ERE HERE ERE HEE ERE HERE ERE EE HE HEE ERE RARER HE HEE ERE HERE HEE REE HE boxplot R Program that will accept an Interpretation and plot a Boxplot Version 1 0 0 Author Thon de Boer Agilent Technologies SiliconGenetics Redwood city CA Last Update 29 Nov 2004 No environment variables defined AAAS AEAEE EEEE EEEE EEE EEE EEEE EEEE EEE EEEE EEEE EEEE EEEE EEEE EEE EEEE EEEE EEEE EEE EEEE EEEE EEEE EEEE EEE EEEE EE HF Get the libraries that are needed library graphics library stats library utils library Biobase library GeneSpring Load the Interpretation i lt GSload int Plot the boxplot of the Normalized data on a logarithmic scale boxplot data frame i a nor names t i expparam 1 log y notc
61. the list of parameters quit if we don t No point doing anything if we don t know which parameter to use i lt charmatch ExperimentParameter row names gs int expparam if is na i q Get the id s for the group and create a sub matrix of only the two group Get the two groups as separate sets and combine them again x lt log2 gs int a nor x cl rep 0 dim x 2 y lt log2 gs int a nor gs int expparam i Condition2 gs int expparam i Conditionl y cl lt rep 1 dim y 2 gs lt data frame x x y y g x cl yel Perform the SAM with the default settings sam out lt sam gs gs cl gene names gs int genenames Create a delta plot if we run this script interactive if interactive delta plot sam out Get the genes that show a Q value less than the Q_cutoff and return the d values geneindex lt sam out q value lt Q_cutoff Create the GeneList genelist lt data frame genes names sam out d geneindex delta sam out d geneindex Write the resulting Gene List GSsave genelist genelist 5 6 Power of ANOVA This script will calculate the power of a gene using the variance within and between group of replicates EHE AE FEAE E AE FE E FE E FE ERE AE FE FE FE AE FE AE RE HEA ERE HERE FE AE AE AE HAE EE AE FE AE HE FE FE AE FEAE FE E FE AE FE E FE E ERE HE AE HE HEE ERE HH FE AE FE AE E AE E AE FE EERE power anova R Program that accepts a GeneSpring
62. the two arguments by clicking on the ADD ARGUMENT button and changing the name of the first argument to CV cutoff and making the default value 100 Press lt RETURN gt after changing the values The second argument name should be Matching conditions with a default value of 1 After the changes have been made the window should look like this GeneSpring GX integration package 25 3 4 5 26 5 Edit External Program Details Program Name Icon Rlogot6 git Program Executable External Program Java Class C HTTP Command Line GS exec R bat cv filter Browse Inputs Outputs Delimiters Arguments User Command Line Arguments Add Argument Argument Name Default CV cutoff 100 Remove Argument Matching conditions 1 Delimiter between argument name and value equals X Fill in missing argument values with There is no need to change the rest of the settings so press SAVE to save the External Program Running the External Program The external program we have just created is stored in the External Programs section in the Navigator window of GeneSpring GX Full Development Hu95aVer2 Genes all genes BAE File Edit View Experiments Colorbar Filtering Tools Annotations Window Help EHA Gene Lists Gene Trees HC Condition Trees Classifications EH Pathways Array Layouts 1 Ho Expression Profiles H External Programs EHO vinsight A Biocondu
63. tings move the following files to the original data directory GS exec H sh boxplot R cv filter Fi DifferentialExpression LPE R gapFilter R power anova R qvalue R SAM R Z score H Moving the files is only required if you have changed your GeneSpring GX data directory 3 4 Creating External Programs using R This section contains a description of the steps that need to be performed or installed to allow GeneSpring GX to execute an R script There are two version of this program one for MS Windows system 65 exec R bat and one for UNIX system called GS exec R sh for UNIX systems 3 44 WINDOWS GS exec R bat This Windows batch execution program GS exec R bat is the main program that will be used to execute the R code This program will take any R script that contains the actual code to be performed and instructs R to execute the program and to send the results back to GeneSpring GX There is usually no need to edit this script since it is a generic script that can work with any proper R script EXCEPT when R was not installed with the installer and the user already has R installed in some different location The contents of the bat file is explained below and although it is not necessary to understand the working of the script it may be useful for future development efforts GeneSpring GX integration package 19 20 The contents of the batch script is presented below echo off set R SCRIPT 1 shift Loop
64. tion procedure see the next section Installation procedure 3 2 4 Upgrading from a previous version of the R integration package This is the second release of the GeneSpring GX R Integration package and if a previous version of the package has been installed most of the files will be overwritten by the new versions of the programs so if any changes have been made that need to be retained make copies of the changes so they could be implemented in the new versions if required 3 2 2 Installation procedure This section describes the steps involved in the installation of the GeneSpring GX R Integration package 3 2 2 1 Installing If you did not install R before or did not use the complete Windows package gs r winall zip you would need to install the R program See the download and installation instructions at http cran us r project org on how to install R for your platform 3 2 2 2 Installing BioConductor The GeneSpring GX R integration package can be used most optimally with the R package BioConductor The BioConductor package is not required for most functionality of the GeneSpring GX R integration package but it is highly recommended to obtain the BioConductor package If you do not already have the BioConductor library and will not install the complete Windows package gs r winall zip you need to install the BioConductor package separately More information about the installation of the BioConductor package can be found at
65. to delete the just downloaded file It is safe to delete the downloaded files since they have been installed and are no longer needed Deleting the files will save disk space GeneSpring GX R integration package EEK R Console DER gt local a lt CRAN packages CRAN getOption E install packages select list a 1 TRUE libPaths i availa trying URL http www bioconductor org bin windows contrib 1 9 PACS Content type text plain length 36778 bytes opened URL downloaded 35Kb trying URL http www bioconductor org bin windows contrib 1 9 GenS Content type application zip length 112094 bytes opened URL downloaded 109Kb package GeneSpring successfully unpacked and MD5 sums checked Delete downloaded files y N y updating HTML package descriptions R 1 9 0 A Language and Environment e For more information about installing packages from BioConductor or R see http www bioconductor org 4 2 GeneSpring functions The GeneSpring package for R contains a number of functions and class definitions that makes it easier to incorporate R scripts into GeneSpring GX The functions will perform the conversion of the specific GeneSpring GX format for Expression Experiments Interpretations and gene lists The package will also provide functions for converting GeneSpring GX objects into BioConductor objects like exprSet and phenoData objects for example The
66. xample of using the Graphical User Interface to R from GeneSpring GX The Rterm program will execute the script that was the first command line argument Now saved into the environment variable SCRIPT to the script The Rterm exe program is located in the rw20170 bin directory in the GeneSpring GX data directory and was installed there by the GeneSpring GX R Integration package installer If you elected to install R as part of the GeneSpring GX Integration package The output from the Rterm program is redirected to a temporary file in the GeneSpring GX data directory to ensure that any error message or warning from the R script will not interfere with the parsing of the results by GeneSpring GX NOTE If you DID NOT use the installer that included the R program and would like to use your own version of R change the path in this script accordingly cat exe GS R out txt This step re directs the output of the R script that is written to a file to the Standard Output stream so GeneSpring GX can pick up this stream and load the data into GeneSpring GX Most R scripts in the integration package will write the results of the analysis to a file called GS out txt in the GeneSpring GX data directory and the cat exe program is used again to re direct the output the STDOUT stream del GS R in txt GS R out txt Rterm output The last step is the clean up stage and the temporary files are deleted using the Windows del command 3 4 1 2 Using the

Download Pdf Manuals

Related Search

Related Contents

(ce) et les objets de pArticipAtioN prÉvus à lA lip  Digrain Aérosol pour Diffuseur + Diffuseur Actif°  User manual Salvus Palermo  protocollo OK verde_Layout 1  Bionaire BWM5905 Humidifier User Manual  Blusens H350-MX  User`s Manual - Floridamusicco.com  Golden Tree Golden Tree (Standard) Ver1.05  9S54 取扱説明書 INSTRUCTION  FD-M785 / FD-M785-E2 FD-M786 / FD-M786-D  

Copyright © All rights reserved. Failed to retrieve file